DigitalTV Transmission

8/7/2019 DigitalTV Transmission

1/37

Dig i t a l TV Transm iss ion:Dig i t a l TV Transm iss ion:Dig i t a l TV Transm iss ion:Dig i t a l TV Transm iss ion:

Channel Coding and ModChanne l Coding and ModChannel Coding and ModChannel Coding and Modulat ionu la t ionu la t ionu la t ionYao Wang

Polytechnic University, Brooklyn, NY11201http://eeweb.poly.edu/~yao


2/37

Yao Wang, 2005 DTV transmission

Out l ine

Overview of ATSC DTV system

Video formats and video coding

Audio formats and audio coding

Channel coding

Modulation

Difference with DVB system


3/37

Yao Wang, 2005 DTV transmission 3

ATSC DTV St andard : Sum m ary

Proved by ATSC (Advanced Television System Committee of

FCC), initially in 1995, Revised Aug 2001 Based on the Grand Alliance HDTV proposal

Video encoding: MPEG2 video (mp@ml for SDTV, mp@hl forHDTV)

Audio encoding: Dolby AC3 (5.1 channel) Multiplex and transport: MPEG2 transport stream syntax

Channel coding and modulation: digital VSB

Terrestrial: 8-level VSB, enabling 19 mbps payload in 6 MHz

channel Cable: 16-level VSB, enabling 38 mbps payload in 6 MHz channel


4/37


ATSC DTV Syst em Model

From Fig.5.1, ATSC DTV standard (A53/B)


5/37


Channel Coding: Overview

To enable detection and correction of bit errors in the received

bit stream CRC: allow error detection only

Ex: adding a parity bit to detect 1 bit error

FEC: allow correction of up to a certain number of errors,depending on the code used and the redundancy introduced

Ex. Repetition code, repeat 3 times of each bit allow correction of 1bit error (1->111,0->000)

Block codes vs. convolutional codes

Channel code rate r= k/n, k information bits, n coded bits

Typical implementation for packetized transmission Source bits + CRC bits -> FEC

If the number of errors exceeds FEC correction limit, CRC candetect the presence of residual erroneous bits, and the packet canbe discarded or otherwise marked as containing wrong bits (w/o

knowing which bits are wrong)


6/37


Conc at enat ed Channel Coding

Outercoder Interleaver

Innercoder Modulation

Channel

DemodulatorInner

decoderDeinterleaver

Outerdecoder

Deinterleaver randomize error bursts due to residual errors afterinner decoder, so that the outer decoder only see randomlydistributed errors


7/37


Why Int er leav ing?

Original symbols:

A,B,C,D, E,F,G,H, I,J,K,L, M,N,O,P,.

After interleaving (depth 4)

A,E,I,M, B,F,J,N, C,G,K,O, D,H,L,P,

Channel error (remaining after inner decoder) affects4 consecutive symbols

A,E,I,M, X,X,X,X, C,G,K,O, D,H,L,P

After deinterleaving

A,X,C,D, E,X,G,H, I,X,K,L, M,X,O,P

Only 1 symbol error in every 4 symbols, can be corrected byan appropriately designed FEC code


8/37


Channel Cod ing in ATSC DTV

Using a concatenated channel coder

An RS outer coder, RS (207,187), symbol length=8. Cancorrect up to 10 symbol errors in 207 received symbols,r=187/207

An interleaver: spread the symbols in each RS block over

many blocks A trellis inner coder (incorporates an inner block interleaver),

r=2/3

For every 2 bits, generates 3 bits, based on the past trace of 2

bits symbols as well as the current 2 bits Total channel code rate = 187/207*2/3=0.60225


9/37


Modulat ion of Dig i t a l Signals : Overv iew

For transmission of digital bits over analog channels

Convert group of digital bits into analog waveforms(symbols)

The analog waveforms are designed according to thedesired carrier frequency

An analog channel of bandwidth B can carry at most 2Bsymbols/s. For reduced inter-symbol interference, lower than2*B symbol rate is used typically

Equalizer is used at the receiver to reduce the inter-symbol

interference (ISI), and multipath effect Revisit the slides for lecture 8 in EE3414


10/37


Am pl i t ude Shi f t K ey ing (ASK )

M-ary ASK: each group of log2M bits generates a symbol. The number corresponding

to the symbol controls the amplitude of a sinusoid waveform. The number of cycles inthe sinusoid waveform depends on the carrier frequency

4-ASK: 2 bits/symbol (00=-3, 01=-1, 11=1, 10=3)

Example: Given a sequence: 01001011, what is the analog form resulting from 4-ASK?

Symbol representation: -1,-3,3,1

Waveform:

00 01 11 10

Symbol interval


11/37


8-ASK

8-ASK: 3 bits/symbol (000=-7, 001=-5, 011=--3, 010=-1, 110=1, 111=3, 101=5, 100=7)

001 011

110 111

010000

101100

The mapping from bits to symbols are done so that adjacent symbols only vary by 1 bit, to

minimize the impact of transmission error (aka Gray coding)


12/37


Quadra tu re Ampl i t ude Modulat ion

(QAM)

M-ary QAM uses symbols corresponding to sinusoids with different amplitude

as well as phase, arranged in the two-dimensional plane.

Ex. 4-QAM (only phase change):

sin(ct)

00=cos(ct-/4)01=cos(ct-3/4)

11=cos(ct-5/4) 10=cos(ct-7/4)

cos(ct)


13/37


Ex am ple of 4 -QAM

01 00 10 11

Example: Given a sequence: 01001011, what is the analog form resulting from 4-ASK?

Using the previous mapping, the analog waveform for the above sequence is


14/37


16-QAM, et c .

16 QAM (4 bits/symbol): 64-QAM (6 bits/symbol)


15/37


Modula t ion in ATSC Syst em : 8-VSB

8-level amplitude shift keying

Each 3 bits generates a pulse, whose amplitude depends on thenumber corresponding to the 3 bits

VSB: retaining only a small portion of the lower sideband, in additionto the upper band The pulse train is multiplied with a sinusoidal wave at the carrier

frequency

A shaping filter is applied to retain only a small portion of the LSB tosave bandwidth

Channel capacity and information rate A 6MHz channel can deliver at most 12 Msymbols/s.

Recall a signal sampled at fs sample/s has a maximum bandwidth of fs/2

But ATSC system uses 10.762 Msymbols/s. With 8-ASK, each symbol carries 3 bits. Thus the date rate is

10.762*3=32.286 Mbps (including channel coding redundancy) Information bit rate = overall data rate * channel code rate

=32.286*0.60225=19.44 Mbps


16/37


By multiplying with a sinusoid signal !

This is known as amplitude modulation (AM)

How do w e shi f t t he f requenc y o f a

s ignal? (Modulat ion Revis i t ed!)

)(tx )cos()()( ttxty c=

frequencycarrier:

signalcarrier

)cos(

c

ct

In 8VSB, an 8-levelpulsed signal


17/37


Frequenc y Dom ain In t erpret a t ion o f

Modula t ion

From Figure

7.5 in[Oppenheim] )(tx

)cos( tc

)cos()()( ttxty c=


18/37


How t o ge t back t o the baseband?

(Demodulat ion)

By multiplying with the same sinusoid + low pass

filtering!

)(ty

)(tw

)cos( tc

m m

)(H

2

)(tx

LPF


19/37


Frequenc y Dom ain In t erpre t a t ion

Figure 7.7 in

[Oppenheim]


20/37


DSB vs SSB vs VSB


21/37


VSB Im plement a t ion

Shaping filter Equalizing filter


22/37


8-VSB Shaping Fi l t er

The shaping filter generally modifies the signal levels.

Each symbol (3 bits) last for a short time, only thevalue at the sampling time at the receiver needs to becorrectly retained.

A Nyquist filter is used as the shaping filter so thatthe signal values at the sampling instances are notmodified.


23/37


Nyquis t Fi l t e r

At any given sampling time (vertical line), only one symbol pulse contributes

to total signal amplitude, all other pulses experience a zero crossing. Theresulting RF envelope corresponds to the eight digital levels only during the

precise instant of sampling.

[From: D. Sparano, What is exactly 8-VSB anyway]


24/37


Effec t o f t he VSB Shaping Fi l t er

Top: Double sideband IF envelope before Nyquist filtering (shaping filter).

Bottom: The same IF signal after Nyquist filtering. The squared-off

transitions are lost and the envelope acquires a noise-like appearance.



25/37


ATSC p i l ot s i gna l

A non-zero DC value (pilot) is added to the

baseband signal (the 8-level pulse signal) beforemodulation to enable carrier extraction at receiver

similar to VSB-C used in NTSC, but uses much lower energy,7% of total


26/37


8-VSB RF Signa l Spec t rum

Pilot(a very low energy carrier signal)

Carrier frequency fc

Only a very smallportion of the LSB

is retained



27/37


Rec eiver Equal izer

Terrestrial broadcasting suffers from multipath effect

Received signal is sum of delayed and attenuated versionsof the same signal with unknown delay variations

Create ghost of the original image

Equalizer = cancellation of inter-symbol interference (ISI)and ghost

Equalizer design: Think of the received signal as the original signal going

through a filter

g(n) = a0 f(n+d0) + a1 f(n+d1)+ a2 f(n+d2)

Can design an inverse filter to recover f(n) from g(n).

Need accurate estimation of aiand di


28/37


29/37


Dat a St ruc t u re

Segmentsync



30/37


31/37


Channe l Coding and Modulat ion in DVB

Also uses a concatenated code

Outer code: RS Inner code: punctured convolutional code

Use very different modulation technique For conversion to complex symbols:

QPSK (=4-QAM, 2 bit/symbol) 16 QAM (4 bit/symbol)

64 QAM (6 bit/symbol)

Resulting complex symbols are split into multiple sub-channels, each using slightly shifted carrier frequency(OFDM)

Coded OFDM: refer to combination of the concatenatedchannel code and OFDM


32/37


Ort hogonal Frequenc y Div is ion

Mult ip lex ing (OFDM)

Basic ideas:

Split the original signal with N samples/s into M sub-channels, each with N/M samples/s and occupying adifferent frequency band (subband decomposition)

The sub-signals must be synchronized so that remultiplexing

is possible at the demodulator The modulation waveforms for the individual sub-channels

should be orthogonal

When using sinusoidal waves with evenly spaced frequencies,can be realized by DFT

Fig.13.34 in Arnold


33/37


Ex am ple w i t h 4 sub-c hanne ls

DEMUX

Freq.

Shift,f0

Freq.

Shift,f1

Freq.

Shift,f2

Freq.

Shift,f3

N sample/s

N/4 sample/s,per sub-channel

System Block Diagram: Original Spectrum

Combined spectrum ofthe 4 sub-channels

From Fig.13.34 in [Arnold]


34/37


Advant age of OFDM

Immunity to fading and interference

Fading and interference in one sub-carrier frequency (sub-signal)does not affect others

Use of interleaving/deinterleaving at multiplexer makes the bursterror on one sub-channel spread randomly over the original signal,which can then be corrected by FEC decoding more effectively.

Also cause reduced interference to adjacent channels

Much better than 8-VSB to dynamic multipath fading (mobilereceivers)

But, compared to 8-VSB

More costly to implement

Require higher transmitter power to achieve the same coveragearea, causing more interference to adjacent analog channels

Lower immunity to impulse noise


35/37


COFDM in DVB

2k mode

2048 sub-carriers/channel, with 1705 of them at center usedfor carrying signals (including pilot signal)

8k mode

8192 sub-carriers/channel, with 6817 used

Each channel is 6, 7, or 8 MHz

DVB-H uses a 4k mode


36/37


What you should k now

Channel coding

Principle of channel coding Benefit of concatenated code and interleaver

Both DTV and DVB uses concatenated channel coding

Modulation:

How to map digital signals to analog waveforms using M-aryASK and M-ary QAM? (illustrate for the case of M=2 or 4)

Principles of VSB and OFDM

DTV uses 8 VSB (8-level ASK and VSB)

DVB uses QAM and COFDM (channel coding plus OFDM)

relative pros and cons of 8-VSB and COFDM


37/37


References

D. Sparano, What is exactly 8-VSB anyway?,http://www.broadcast.net/~sbe1/8vsb/8vsb.htm

8-VSB, http://en.wikipedia.org/wiki/8-VSB, Oct 2005

J. H. Stott, The how and why of COFDM,

http://www.ebu.ch/en/technical/trev/trev_278-stott.pdf FCC/OET, REPORT ON COFDM AND 8-VSB

PERFORMANCE, Sept. 1999

http://www.fcc.gov/Bureaus/Engineering_Technology/Documents/reports/dtvreprt.pdf